Heatguard Edge
“HeatGuard Edge doesn't just measure heat—it predicts heat stress before it becomes an emergency, helping protect lives through intelligent, real-time Edge AI.”
- Problem, approach and result explained
- Expert reviewed· 5 reviews
Overview
The problem
-
Who has the problem? Outdoor workers, farmers and construction workers exposed to extreme heat.
-
Why does it matter? Prolonged heat exposure can cause heat stress, dehydration and serious health emergencies.
-
What did we set out to do? To develop HeatGuard Edge, an Arduino UNO Q-based system that uses multiple sensors, WBGT calculation and Edge AI/ML (GRU and anomaly detection) to monitor environmental conditions, predict heat risks and provide timely alerts through a local dashboard.
The approach
that fought u the most ?
- Components and why? BME688: Measures air temperature, humidity and VOC.
DS18B20: Measures radiant globe temperature (in our globe setup).
MLX90614: Measures surface temperature without contact.
BH1750: Measures ambient light intensity.
Anemometer: Measures air speed.
ESP32: Collects sensor readings and handles node-level processing.
LoRa: Transmits readings wirelessly from sensor nodes to the central system.
Arduino UNO Q: Central processing, local dashboard and Edge AI inference.
OLED: Displays WBGT and heat-risk status at individual nodes.
Solar panel + LiFePO4 battery: Provides power for field deployment.
- How is it wired? Sensors connect to ESP32 through interfaces such as I2C, One-Wire and ADC, depending on the sensor.
ESP32 communicates with the LoRa module through SPI.
LoRa transmits sensor data wirelessly to the central receiving unit.
The receiving system forwards readings to Arduino UNO Q.
UNO Q's Linux processor runs the Python backend, dashboard and ML models.
OLED displays receive the relevant readings and risk status through their respective node controllers.
- What part challenged us the most? Integrating hardware, communication and Edge AI into one working system. Understanding how multiple sensors collect data, how ESP32 transmits it through LoRa, and how Arduino UNO Q processes those readings to calculate WBGT, run GRU forecasting and display real-time risk predictions locally was the most challenging part.
The result
- Does it work in the real world?
Yes, our prototype demonstrates real-time environmental monitoring, WBGT calculation, heat-risk assessment and local Edge AI forecasting. However, it still needs extensive field testing before real-world deployment.
- What specifications did we achieve?
- Multi-sensor environmental monitoring.
- Three distributed sensor nodes.
- Wireless ESP32–LoRa communication.
- Local processing using Arduino UNO Q.
- WBGT calculation and GRU-based forecasting.
- Anomaly detection and local dashboard.
- OLED-based WBGT and heat-risk display at each node.
- How did we test it?
- Tested individual sensors against their expected measurement ranges.
- Checked ESP32 sensor-data acquisition.
- Verified wireless communication between nodes and the central system.
- Compared calculated WBGT against expected readings under different environmental conditions.
- Tested ML forecasting and anomaly detection using collected/sample datasets.
- Verified local dashboard updates and OLED risk displays.
Reflection
what did u learn ?
- What would we do differently?
Improve sensor calibration and measurement accuracy.
Use more reliable long-range communication.
Collect more real-world data to improve ML model accuracy.
Optimise power consumption for continuous outdoor operation.
Submitted to Heat Intelligence at the Edge
- Arduino board used?
Yes
- Edge AI used?
Yes
- AI/ML model used
GRU (Gates Recurrent Unit)
- Sensors used
BME688 DS18B20 MLX90614 BH1750
- Camera used
MLX90614 (Thermal Camera)
- Modulino used
None
- Problem being solved
HeatGuard Edge is a low-cost, real-time, edge-AI based heat-stress monitoring and response system that continuously senses environmental conditions, calculates heat risk locally, predicts upcoming conditions, detects anomalies, and triggers alerts/actions before the situation becomes dangerous.
- AI model training platform
PyTorch
- Arduino UNO Q features used
edge AI , the ML models stays at Arduino's QRB2210 MPU instead of cloud
- Deployment method
locally it works , in uno Q itself
- Key innovation
Localized, predictive heat-stress monitoring with autonomous response using edge AI.
- Solution summary
HeatGuard Edge is an edge-AI-powered heat-stress monitoring and prevention system that measures localized environmental conditions, estimates heat risk and predicts future heat stress. Using multi-sensor ESP32–LoRa nodes and the Arduino UNO Q, it processes data locally, provides real-time alerts and can automatically activate connected cooling systems. Solar-powered backup and offline operation support resilience during power and network disruptions, while a dashboard enables monitoring of multiple locations.
- Report (PDF)
- Report (PDF)PDF
- What's working now?
1. Who has the problem? Outdoor workers, farmers and construction workers exposed to extreme heat. 2. Why does it matter? Prolonged heat exposure can cause heat stress, dehydration and serious health emergencies. 3. What did we set out to do? To develop HeatGuard Edge, an Arduino UNO Q-based system that uses multiple sensors, WBGT calculation and Edge AI/ML (GRU and anomaly detection) to monitor environmental conditions, predict heat risks and provide timely alerts through a local dashboard.